Add Generation.text error log

[dinggege1024]

you could trace this background https://github.com/hwchase17/langchain/issues/3654

[dinggege1024]

@hwchase17 Please review this pull request and provide some suggestions, tks

[vowelparrot]

I'd prefer to either/both:

• Handle this upstream -> if the response of _generation isn't a string, then the typing is wrong and we should fix that in the LLM wrapper
• Handle this in the Generation model itself -> if the validation error message isn't helpful then we can do so there

Are there examples where this is occuring?

[dinggege1024]

I'd prefer to either/both:

• Handle this upstream -> if the response of _generation isn't a string, then the typing is wrong and we should fix that in the LLM wrapper
• Handle this in the Generation model itself -> if the validation error message isn't helpful then we can do so there

Are there examples where this is occuring?

I used LLM with our api, i goes wrong with response in above pic. Before debug, i can not understand what happen.

-- Handle this in the Generation model itself -> if the validation error message isn't helpful then we can do so there Updated pr, Is the solution mentioned above? Tks

@vowelparrot

[hwchase17]

agree with @vowelparrot this probably isnt the best place to handle it

I used LLM with our api, i goes wrong with response in above pic theres probably some misconfiguration of your LLM wrapper going on, i would fix that

[dinggege1024]

agree with @vowelparrot this probably isnt the best place to handle it

I used LLM with our api, i goes wrong with response in above pic theres probably some misconfiguration of your LLM wrapper going on, i would fix that

Thks for reply, if fixed up please show me the way~

[baskaryan]

stale, closing but let me know if you're still seeing issues!